Error on path analysis using SEM

[SRL]

I have been doing path analysis using stata sem.
The dependent variable is not a latent variable, and I have not had any problems until today.
For some reason, i am getting an error message and stata not running the path analysis because my dependent variable is not a latent variable.
Any suggestions or solutions?

[Clyde Schechter]

I think your chances of getting helpful advice will improve if you show us exactly the commands you gave and exactly the response you got from Stata. Do this by re-running the command and then pasting the echoed command plus Stata's output into a code block. (See the Forum's FAQ for how to set up a code block.)

Also, if you click on CONTACT US you will be able to request a change of your username to your full first and last name, helping us maintain the professional atmosphere on the forum.

[ben earnhart]

I don't know what you were doing originally, but one solution is to declare a latent dependent variable with a loading constrained to 1 and error variance constrained to zero. Thus, the latent is exactly the observed. This is how I've had to deal with similar situations, dunno how you had a different work-around before.

[Clyde Schechter]

@Ben. Yes that's a workaround. But no workaround should be needed. I've run plenty of path analyses. In fact, I probably use SEM for that more than I do for things involving latent variables. And I've never encountered that problem before. I mean, just try -sem price mpg foreign- on the auto.dta set: it works with no complaint. So I suspect that the original poster has either done something syntactically wrong, or is misunderstanding an error message, etc. That's why I think he/she should show us exactly what was commanded and what the response was.